So you doom OTP on one account, and are more flexible on another and wonder why they might be different ranks…
If I start an account and otp Ball. I’ll hit diamond/masters fairly efficiently.
If I do the same and otp doom, or Genji, I’ll barely get out of gold.
I play the same heroes on each one. If you wana be a Doom specialist your gona need multiple accounts trust me.
I decided to otp Doom on this old account. Open profile btw. Im 25hrs in around 120 matches and only at plat 5. Im doing it to see exactly how long it will take just to hit D5. Stomping every game win or lose.
My plan is to make a brand spanking new account and see how fast I get it to where I belong.
Im not here to argue with you, infact I agree with most things you say about the game. I will never say a person cant climb and the game does not forces losses. All im saying is something wild is happening with the matchmaker and mmr.
I will post my results when I get this account to D5 and then the new one. To compare.
Most importantly though is playing for fun. This game is fun to me no matter how frustrating it gets
New accounts take your QP mmr into account now, which may place you very fast to some rank in comp. Though supposedly they lowered its impact a bit recently (Gavin said something pointing to that)
But new accounts also have normally a lot of uncertainty factor on them, and win (and loss) streaks have affected them a lot. Kind of what can happen with placements on old accounts in this season, especially the higher rated they were.
Anyways, overwatch rating is more about long time averages really, at least to me.
How fast you can hit a “peak” is not necessarily that telling. I wish Ow did show a average rating you had in the game for the whole season or so.
Many people do however create an alt, hit a new peak and then stop playing that alt (rank camping).
Had they played the alt as long as their old account, it’s very possible they would average out to the same rating as on their main account.
In the end you have to ask yourself why you’re playing the comp mode and what you’re looking from it. Just hitting a new high peak as fast as possible or getting consistent?
yup, they force you with bad teamates, and then punish you for having a loss streak xd
also i cant believe they punish you the same as a normal loss when you have a leaver.
How can they, then the only think they use to match you in a single number…
Yeah part of their game design is to force losses, something about playing longer and being online more , when you lose more than win.
That isn’t a thing. It’s a conspiracy spouted but a lot of games. But isn’t a thing.
Everyone i know will stop playing on loss streaks rather than on winning. Maybe I and everyone I know are the odd ones though.
Besides that, the game is not forcing anything on an individual player.
Here’s also a funny thing, you cant make everyone lose more than they win. That would mean everyone would have a negative win rate - which is impossible.
Scott Mercer designed the matchmaker. Watch his interviews.
He has stated that if the matchmaker is working as intended, your win rate will be around 50%.
He has also stated there are matches where you are supposed to win and supposed to lose.
It’s not a stretch/conspiracy for someone to try and figure out how these 2 points works together and come to the conclusion of “forced” 50% winrate or “forced” wins/losses.
Yes, this has been apparent for years regardless of the modifiers. The more you win, the more team mates you’re placed with that are losing. The more you lose, the more team mates you’re placed with that are winning. That’s why effort is pointless unless the game is so close that you can be the difference maker (few and far between outside of smurf situations).
Matchmaking favors MMR heavily when assigning teams.
No, it’s not. Not when it’s sometimes. When it’s significantly rare then it’s okay. Don’t caress the dev’s babymakers. That’s terrible coding. Extremes happening often are lousy coding. The game needs to properly gauge skill and the w/l only way isn’t doing it properly as 1 in 10 of my games could go either way, the rest are stomped or be stomped.
OW couldn’t create a decent ranking system, so it hired a third party to manage its competitive circuit because its own system wasn’t competitive.
If they could apply ML here, it would be better than the garbage coding they are using currently.
How would this work, because every player should be then treated the same.
So:
If you are winning, then all the 4 other players in your team must have been losing, to be put into your team because you have been winning.
But if your teammates were losing, they should be put into games with players who were winning?
Why would they be put into team with others who are also losing?
Is it only about that one player that it cares about for some reason and not the rest?
MMR is the only factor used. They dont look into past match history.
I must admit, if Small package is right, and thats a big if, coming from me… having modifiers based on SR would be the gap in “real” machine learning. Like why train the model on something “indirect” of the model and not the model itself?
Meaning, if the machine was wrong on the MMR side of things, why increase SR? Meanwhile at the same time your SR was Low compared to your MMR, so in theory your rewarding the player for an “expected win”…
You have to overlay the -6 6 range with the same SR range and you will see each mmr value has a range of SR. Rewarding off of SR creates inflexion points where your almost out of a mmr sub range.
Here’s what Morgan has to say:
off topic, I’ve thought of setting up an ow2 mmr faq website just for the heck of it.
Every player is treated the same. High MMR is matched with low MMR to “balance” the team.
Not necessarily, but the more you win, the higher your MMR goes, meaning the lower average MMR your team mates will have.
Strictly wrong. Rank is used to capture a baseline group, then MMR searching is applied.
Ok - thats different to saying that you are put into games with players who have been losing. They may have lower mmr but still been winning.
Edit: Also if you have been winning and your mmr has gone up, unless you’re at the top of the whole ladder, you will be in games where your mmr may in fact be in the low side of the lobby.
This is why they need to dump the MMR and leave it as just SR across the board. Your placements should be tracking more than just W/L given that they have the data for your average player on every hero at each rank. That should then place you in the proper rank to start from then SR should be using W/L plus stats to move players up/down quicker.
Yes, SR “follows” your MMR, not saying it doesnt, but math prevails, -6 to 6, overlay with 0 to 5000k mmr accounts for 200 points of SR range. There is a little wiggle room.
They’re still losing more than you, hence
It’s not an absolute black & white thing, it’s a chaotic scale that favors the MMR within a bracket. And a player’s MMR is extremely volatile, based on their recent W/L record. So by that metric, if you’re winning more games than losing, you will be put with people who are losing more than winning. That’s how the math in the matchmaker works.
A balanced match will be about 50% chance of winning. If the match maker works as intended, most (if not all) your games will be like this. Thus meaning a 50% winrate is highly likely. Not forced, just a mathematical probability.
He’s not said that, he’s said that as you move about MMR and because queues are are queues, that 50% change of win, might mathematically be 55% or 45%. This is where the modifiers come in. If the SR is slightly lower on one team, but they win, they will get the relevant modifier.
It is more that people have interpreted things they way they want, rather than the way they were. If someone believes the theory, they will find anything they can and call it “evidence”.
That isn’t true. It is more coincidence. If you win a lot and your MMR goes from 0.85 to 1.25 (for example). You will be matched with 9 others at 1.25. Some of them are possibly going to be on a loss streak and dropped a bit, others will be on a win streak. And just as likely you’ll have some who are about 50/50.
Streaks/recent results aren’t relevant to match making. It only considers 4 things
As a core, matchmaker is a mathematical formula that takes into account multiple factors, including but not limited to:
- Skill: The matchmaking system will try to group players with and against players of similar skill. The game represents “skill” as a hidden matchmaking rating, or “MMR.” In role queues, the matchmaker will prioritize placing pairs of similar MMR players on each role on either team. All game modes, including, only look at matchmaking rating when creating a match; outward-facing skill tier and division have no effect on matchmaking.
- Queue time: The matchmaking system will try to find a fair match in as little time as possible.
- Ping: Players connected to servers that are geographically closer to one another are prioritized in order to minimize lag.
- Group size: The matchmaking system will attempt to find groups of equal sizes to match against each other. For example, if four players are looking for a game together, the matchmaker will try to match them against another group of four.